Michael Lattke1, Robert Goldstone2, James Ellis3, Stefan Boeing4, Jeronimo Jurado-Arjona5, Nicolás Marichal5, James MacRae3, Benedikt Berninger5,6,7,8, Francois Guillemot1*
1Neural Stem Cell Biology Laboratory, The Francis Crick Institute, London, NW1 1AT, UK 2Advanced Sequencing Facility, The Francis Crick Institute, London, NW1 1AT, UK 3Metabolomics Facility, The Francis Crick Institute, London, NW1 1AT, UK 4Bionformatics & Biostatistics, The Francis Crick Institute, London, NW1 1AT, UK 5Institute of Psychiatry, Psychology & Neuroscience, Centre for Developmental Neurobiology, King's College London, London, SE1 1UL, UK 6MRC Centre for Neurodevelopmental Disorders, King's College London, London, SE1 1UL, UK
7Institute of Physiological Chemistry, University Medical Center of the Johannes Gutenberg University Mainz, 55128 Mainz, Germany 8The Francis Crick Institute, London, NW1 1AT, UK
*lead author
Corresponding author e-mail address: Francois.Guillemot@crick.ac.uk
**Contacts for this online resource:** Stefan.Boeing@crick.ac.uk, Michael.Lattke@crick.ac.uk
Astrocytes have essential functions in brain homeostasis that are established late in differentiation, but the mechanisms underlying the functional maturation of astrocytes are not well understood. Here we identified extensive transcriptional changes that occur during murine astrocyte maturation in vivo and are accompanied by chromatin remodelling at enhancer elements. Investigating astrocyte maturation in a cell culture model revealed that in vitro-differentiated astrocytes lacked expression of many mature astrocyte-specific genes, including genes for the transcription factors Rorb, Dbx2, Lhx2 and Fezf2. Forced expression of these factors in vitro induced distinct sets of mature astrocytes-specific transcripts. Culturing astrocytes in a three-dimensional matrix containing FGF2 induced expression of Rorb, Dbx2 and Lhx2 and improved astrocyte maturity based on transcriptional and chromatin profiles. Therefore, extrinsic signals orchestrate the expression of multiple intrinsic regulators, which in turn induce in a modular manner the transcriptional and chromatin changes underlying astrocyte maturation.
Online Resource Version 0.1.4 (21-05-18)
Rationale Astrocytes have essential functions during brain development and homeostasis. Immature astrocytes in the developing brain support developmental processes such as neuronal maturation, while mature astrocytes in the adult brain support brain homeostasis and modulate neuronal signalling. However, the stages of astrocyte development and the molecular changes underlying these functional changes are not well understood. Here we performed scRNA-Seq to characterise the transcriptional changes along the astrocytic lineage trajectory in the murine striatum.
Experimental design Astrocytes were acutely purified from striatum tissue at postnatal day 3 (P3, for immature astrocytes) or at 3 months of age (3m, for mature astrocytes) using ACSA2-MACS. Astrocytes from 3 independent preparations (each 2-5 animals) were analysed using the 10X-Genomics Chromium scRNA-Seq system.
Analysis outline Cells with >500 features (detected genes) and < 10% mitochondrial content were clustered using Seurat3 (Figure 1C). Sox9-postive astroglial lineage clusters were isolated and reclustered (Figure 1D). Pseudotime analysis was performed using Slingshot (Figure 1E). Genes with dynamic expression along pseudotime were identified using tradeSeq, and grouped according to their mean expression in pseudotime bins (Figure 1F).
Results summary In the Seurat analysis we identified several clusters of transcriptionally clearly distinct immature and mature astrocytes. Slingshot pseudotime analysis identified a likely maturation trajectory which includes two populations of proliferating progenitors (clusters 6, 8), two consecutive stages of postmitotic immature astrocytes (clusters 2, 0), and two main populations of mature adult astrocytes (clusters 7, 14). Our pseudotemporal gene expression analysis revealed 4 groups of immature astrocyte genes and two groups of mature astrocyte genes with distinct expression patterns.
Figure 1A: scRNA-Seq of striatal astrocytes; UMAP plot showing clustering of all analysed cells and expression of the astroglial lineage marker Sox9 (related to manuscript Figure 1C)
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 1B: UMAP plot showing reclustering of Sox9+ astroglial cells, developmental stage and expression of the proliferation marker Mki67 (indicating putative glial progenitors); (related to manuscript Figure 1D)
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 1B: Clusters (Sox9+ cells reclustered)
Figure 1C: Slingshot pseudotime analysis of the astroglial lineage; PCA plot showing Sox9+ astroglial cell clusters, and pseudotime along the putative main striatal astrocyte lineage; (related to manuscript Figure 1E)
Cluster labels from the Seurat reclustering (Figure 1D) were used to infer lineage relationships with the Slingshot algorithm (using Cluster 6 as starting point (Proliferation-marker expressing progenitors with highest negative PC1 loading)). Lineage 6, which includes the main adult astrocyte clusters 7 and 14 (highest positive PC1 loading), was considered as the main striatal astrocyte differentiation/maturation trajectory.
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 1C: PCA plot with clusters (Sox9+ cells reclustered)
Figure 1C: PCA plot with pseudotime (Sox9+ cells reclustered)
Figure 1D: Gene expression changes along pseudotime (main striatal astrocyte lineage) (related to manuscript Figure 1F)
Mean gene expression of all cells in each pseudotime bin (equal pseudotime ranges) was calculated and centered on the mean of all bins. Bins 1 and 20 were omitted as they contained less than 10 cells each. Genes shown were identified as dynamically regulated by the tradeSeq algorithm. Genes were further filtered to remove lowly changing genes with less than 0.2 deviation from the mean in all bins.
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
[heatmap view under construction]
Pseudotime and pseudotime bins by cluster (Sox9+ cells reclustered)
Rationale We analysed maturation of cortical Astrocytes in vivo using bulk RNA-Seq and compared the differentially expressed genes with the immature and mature astrocyte-specific genes identified in the striatal scRNA-Seq analysis, to validate the scRNA-Seq analysis and to identify transcriptional changes that are common during astrocyte maturation in different brain regions.
Experimental design Astrocytes were acutely purified from cortex tissue at postnatal day 4 (P4, for immature astrocytes) or at 2 months of age (2m, for mature astrocytes) using ACSA2-MACS. Astrocytes from 3 independent preparations (each 2-5 animals) were analysed using bulk RNA-Seq.
Analysis outline DESeq2 was used to identify differentially expressed genes.
Results summary Our analysis identified a high confidence signature of immature and mature astrocyte-specific genes which are changing during maturation in both the striatum and cortex, independent of the different analysis approaches.
Figure 2: Volcano plot showing gene expression changes in cortical astrocytes from P4 to 2m (related to manuscript Figure 2B)
Genes enriched in cortical astrocytes at P4 (left) and at 2 months of age (right). Maturation-regulated genes from the striatal scRNA-Seq analysis are highlighted.
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 2: Volcano plot with immature genes from striatal scRNA-Seq analysis
Figure 2: Volcano plot with mature genes from striatal scRNA-Seq analysis
[under construction]
Figure 3A: Expression of transcription factors that might bind motifs enriched in chromatin changing accessibility (bulk RNA-Seq of cortical astrocytes (data from Figure 2) (related to manuscript Figure 3G)
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 3A: ETS transcription factors enriched in immature astrocytes
Figure 3A: ROR and HOX transcription factors enriched in mature astrocytes
Rationale Astrocyte differentiation in vitro was analysed by bulk RNA-Seq to assess to which extent maturation is recapitulated in vitro and to characterise this model for the further analysis of the mechanisms underlying astrocyte maturation. RNA-Seq data from the in vitro model were compared to data from cortical astrocytes in vivo (from Figure 2).
Experimental design Neural stem/progenitor cells cultured in EGF/FGF2-containing medium and astrocytes differentiated from these cells by growth factor withdrawal and exposure to BMP4-containing medium for 14 days were analysed by RNA-Seq.
Analysis outline DESEq2 was used to indentify differentially expressed genes. Gene expression was compared to immature and mature cortical astrocyte profiles from above, focussing on the common immature/mature astrocyte-enriched genes from the cortical and striatal in vivo analyses (see above, Figures 1/2).
Results summary We show here that astrocytes in our in vitro model fail to induce a large part of the mature astrocyte gene signature, suggesting that they remain partially immature. Among these genes are several transcription factors, whose lack of expression might explain the lack of maturation in vitro.
Figure 4A: Expression of selected mature genes in vivo vs in vitro. (related to manuscript Figure 4C)
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 4A: Expression of selected mature astrocyte genes in BMP astrocytes
Figure 4B: Expression of selected mature transcription factors in vivo vs in vitro. (related to manuscript Figure 4G)
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 4B: Expression of selected mature astrocyte transcription factors in BMP astrocytes
Rationale We investigated the role of individual transcription factors in astrocyte maturation, focussing on four candidate maturation regulators Rorb, Dbx2, Lhx2 and Fezf2, which were not induced during astrocyte differentiation in our in vitro model. In a reconstitution assay, we forced expression of these factors in vitro and analysed the consequences for the expression of mature astrocyte genes by RNA-Seq.
Experimental design Astrocyte were differentiated from culutred neural stem cells by growth factor withdrawal and exposure to BMP4 containing medium. At day 6 of differentiation, astrocytes were infected with a lentiviral vector to induce expression of the candidate regulators Rorb, Dbx2, Lhx2, Fezf2 or an EGFP control transgene. RNA-Seq was performed at day 14.
Analysis outline DESEq2 was used to indentify differentially expressed genes. Gene expression was compared to immature and mature cortical astrocyte profiles from above, focussing on the common immature/mature astrocyte-enriched genes from the cortical and striatal in vivo analyses, which lacked expression in EGFP controls in vitro (see above, Figures 1/2/4).
Results summary We could show that the expression of Rorb, Dbx2, Lhx2 and Fezf2 in vitro induced distinct subsets of mature astrocyte genes, suggesting a modular control of maturation by multiple transcription factors.
Figure 5: Selected mature genes induced by transcription factor expression in vitro (related to manuscript Figure 5C)
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 5: Selected mature genes induced by candidate transcription factors
Rationale The transcription factors Rorb and Fezf2 induced distinct subsets of mature astrocyte genes in vitro. To assess whether both factors directly cooperate in induce astrocyte maturation, we expressed both factors in combination in our in vitro model.
Experimental design Astrocyte were differentiated from cultured neural stem cells by growth factor withdrawal and exposure to BMP4 containing medium. At day 6 of differentiation, astrocytes were infected with lentiviral vectors to induce expression of the Rorb and Fezf2 in the same cells. RNA-Seq was performed at day 14.
Analysis outline DESEq2 was used to indentify differentially expressed genes. Gene expression was compared to immature and mature cortical astrocyte profiles and in vitro astrocytes expressing EGFP or Rorb or Fezf2 individually (data from Figure 5), focussing on the common immature/mature astrocyte-enriched genes from the cortical and striatal in vivo analyses, which lacked expression in EGFP controls in vitro (see above, Figures 1/2/4/5). Genes displayed in the figure were induced only by Rorb and Fezf2 in combination.
Results summary We could show that the expression of Rorb and Fezf2 in vitro induced additional mature astrocyte genes not induced by the individual factors, suggesting a cooperative synergistic activity of both factors in astrocyte maturation.
Figure 6: Mature genes induced only by combined expression of Rorb and Fezf2 (related to manuscript Figure 6D)
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 6: Mature genes induced only by combined expression of Rorb and Fezf2
Rationale Astrocytes in our in vitro model fail to induce many genes induced by maturation in vivo, including transcription factors like Rorb, Dbx2, Lhx2 and Fezf2, which promoted maturation in our previous experiments. We hypothesised that this might be due to a lack of extrinsic maturation signals missing in culture. We therefore assessed whether signals that have been reported to promote functional maturation (FGF2 and increased cell-cell-contacts in three-dimensional cultures), are able to induce transcription factors not induced in conventional cultures, and whether this results in improved transcriptional maturation.
Experimental design Astrocyte were differentiated from cultured neural stem cells by growth factor withdrawal and exposure to BMP4 containing medium for 7 days in 2D or 3D culture conditions. Astrocytes were then matured in basal or FGF2 containing medium for 7 more days. RNA-Seq was performed at day 14.
Analysis outline DESEq2 was used to indentify differentially expressed genes. Gene expression was compared to immature and mature cortical astrocyte profiles (data from above), focussing on the common immature/mature astrocyte-enriched genes from the cortical and striatal in vivo analyses, which were not induced in basal 2D conditions (see above, Figures 1/2/4).
Results summary We could show that the FGF2 and 3D culture conditions both induced mature transcription factors not induced in control cultures differentiated in basal medium in 2D. In particular, FGF2 in 3D conditions induced Rorb, Dbx2 and Lhx2 and promoted an overall more mature transcriptional profile. This suggests that indeed extrinsic maturation signals lacking in conventional cultures are required for astrocyte maturation.
Figure 7A: Selected mature transcription factors induced by maturation signals (related to manuscript Figure 7B)
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 7B: Selected mature genes induced by maturation signals (related to manuscript Figure 7C)
Click on the links below to view the figures and the underlying dataset in the interactive viewer:
Figure 7B: Selected mature genes induced by extrinsic signals/culture conditions